Introduction

Globally, more than half of children between the ages of 2 and 17 years reported having experienced some type of violence each year, according to a systematic review of global violence prevalence [1]. Africa bears the biggest brunt of violence against children (VAC), with the prevalence of physical violence at 60% and 51% for boys and girls, respectively [2]. Violence is endemic to South Africa, particularly associated with structural and systemic violence originating from the Apartheid period. A nationally representative survey among children 15–17 years of age reported that almost 20% of children had experienced some form of sexual abuse, 21.3% experienced neglect, and over 30% and 16% experienced physical and emotional abuse, respectively [3]. Children are likely to experience different forms of violence at different stages of their lives, as their environments and interactions change [4]. This ranges from violent discipline during early childhood by caregivers, bullying by peers at school, sexual violence victimization and intimate partner violence during the adolescent period – although the boundaries of these timeframes are porous [5,6]. The prospective Birth to Twenty Plus study in Soweto found that almost 40% of children were exposed to up to six types of violence throughout their lives, including exposure to violence in the community, home and school; exposure to peer violence; being a victim of physical violence, and sexual violence [7].

Despite mounting evidence of the determinants and the burden of VAC, few studies have demonstrated the effects of life course exposure to multiple forms of violence, and even fewer studies have identified the different clusters of violence patterns (trajectories) over time. Previous studies, conducted outside of Africa, have demonstrated heterogeneity in the number and shapes of trajectories of physical and/ or sexual violence victimization, with two [8,9] three [8,10], four [11] and five [12] trajectory groups identified. Most of these studies assess trajectories of violence victimization as predictors of subsequent physical and mental health outcomes in adulthood, rather than the identification of risk factors for trajectory group membership. And where they do, most studies assessing risk factors for violence trajectory group membership in childhood tend to focus on violent aggression or perpetration. However, a significant overlap between victimization and perpetration experiences is documented in research [13–16], showing that victims and perpetrators share similar contextual features that place them at an increased risk for violence victimization and perpetration.

The developmental origins of health and disease (DOHaD) perspective stipulates that experiences or exposures during sensitive developmental period – specifically during utero and early childhood – can have negative long-term consequences in the risk for poor health, disease and behavioural problems throughout the life-course [17,18]. Several demographic and early life factors have been associated with violence trajectories. Being male, residing in households with low socioeconomic status, early motherhood, having more than one child in the household, single parenthood and low parental education were some of the risk factors identified for membership in persistent physical violence victimization or aggression trajectory groups [11,19–21]. From cross-sectional analyses conducted in different countries, early life factors predicting violence victimization include household crowding [22], paternal absence [23] and mother’s prior history of violence [24,25]. Factors along the pathway of violence experiences have also been identified. Both low [26] and high birth weight [27,28], and catch-up growth (weight and length) in infancy are documented to predict future overweight and obesity [29,30], delayed cognitive development [31] and poor academic performance [29], which are risk factors for violence victimization and perpetration [32–34].

There is a dearth of studies conducted in Africa assessing factors associated with childhood trajectories of violence due to a lack of longitudinal data on violence. This limits the ability to assess the developmental course of violence and the heterogeneity of violence trajectories in populations outside of the Global North. This is paramount towards the identification, development and implementation of context-specific prevention strategies prior to the child’s exposure or experience of violence. The current study aims to generate physical and sexual violence victimization trajectories across childhood and assess which factors in early life predict membership into specific violence trajectories.

Materials and methods

Study design

This study involved secondary data analysis of the prospective Bt20 + cohort study. Data from birth until 18 years (1990–2018) from the primary study were included in this secondary study. The study design is illustrated in Fig 1 below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study design.

https://doi.org/10.1371/journal.pone.0294207.g001

Results

Study characteristics

A total of 3,273 children were enrolled at birth. Of these, 2,057 (62.9%) reported experiences of physical violence victimization and 2,051 (62.7%) reported sexual violence victimization in at least two of the four time points of data collection, and were included in the study. Experience of physical violence ranged from 14.0% at age five, with a peak of 71.8% at age 15, which declined to 38.6% at 18 years of age. Sexual violence experiences ranged from 0.8% at five years to a peak of 28.9% at the age of 18 years (Table in S1 Table). A total of 1,212 (37.0%) and 1,218 (37.2%) participants did not meet the inclusion criteria and were excluded from the analyses for physical and sexual violence victimization trajectories, respectively. For both analyses, there were no significant differences between included and excluded samples with reference to sex, maternal age at birth, father’s presence and birth weight. Compared to children excluded in this study, a greater proportion of children included in the study resided in poorer households, which were crowded and headed by a single parent. In addition, the included sample had a lower proportion of parents with post school training compared to children in the excluded sample (Tables in S2 and S3 Table). Covariates for physical and sexual violence victimization trajectories that were significantly different between included and excluded samples, were adjusted for in the multivariable analyses.

Study characteristics by sex.

Among the 2,057 children included in the physical violence and 2,051 children included in the sexual violence victimization trajectory analyses, there was a higher proportion of girls (52.5% physical violence; 52.4% sexual violence) than boys (Table 1). More girls (12.7%) than boys (9.7%) were born with low birth weight. Overall, a higher proportion of children were born in households with low socioeconomic status (65.0% physical violence; 64.8% sexual violence), had one or more siblings (61.9% physical violence; 61.8% sexual violence) and resided in a single parent household at birth (62.0%). A higher proportion of fathers (18.8% physical violence; 18.7% sexual violence) compared to mothers (8.6%) had attained higher schooling after completion of secondary school in both analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Description of study sample characteristics stratified by sex.

https://doi.org/10.1371/journal.pone.0294207.t001

Trajectories of physical and sexual violence victimization

Following the examination of different trajectory groups and polynomial functions, a two-group model was selected as the most parsimonious model that best described patterns for both physical and sexual violence victimization between five and 18 years of age. However, the shape or patterns of violence victimization differed between physical and sexual violence victimization trajectories.

Physical violence victimization trajectories.

Both group trajectories of physical violence victimization had an early onset of violence victimization. These groups were (a) adolescent limited (65.1%) characterized by a gradual increase in violence victimization from age five to 15 years, after which there was a decrease in violence victimization, and (b) the chronic increasing group (34.9%) characterized by a persistently increasing pattern of violence victimization throughout adolescence (Fig 2). The two-group model selected had the lowest BIC value, with average posterior probabilities for group membership greater than 0.70 for the adolescent limited group (0.93) and the chronic increasing group (0.73). This model had medium entropy of 0.56 reported (Table in S4 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Physical violence victimization trajectories among children between 5 and 18 years of age.

Note; lines show estimated trajectories, points at each age represent the observed proportion of children between the age of 5 to 18 years reporting physical violence victimization according to the assigned physical violence trajectory group.

https://doi.org/10.1371/journal.pone.0294207.g002

Sexual violence victimization trajectories.

Fig 3 below shows the two-group trajectories of sexual violence victimization that were characterized by a late onset of sexual violence victimization. These groups were the (a) adolescent limited (74.1%) and (b) late increasing (25.9%) trajectory groups of sexual violence victimization. The adolescent limited group was characterized by low sexual violence victimization at the age of five years, whose onset was at age 11 with a peak at age 15. This was followed by a decrease in experience of sexual violence from age 15. The late increasing trajectory group also showed onset at age 11, with sexual violence victimization persisting and increasing through to age 18. The model with a quadratic (2) and cubic (3) term provided the best description of sexual violence victimization trajectories. Although the 2-group model with quadratic terms had a slightly lower BIC compared to the model selected, the significantly lower entropy (0.44) for that model made it less desirable for describing sexual violence victimization patterns. The final model selected had average posterior probabilities of 0.85 and 1.00 for the adolescent limited and late increasing group, respectively, with an entropy of 0.57 (Table in S4 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Sexual violence victimization trajectories among children between 5 and 18 years of age.

Note; lines show estimated trajectories, points at each age represent the observed proportion of children between the age of 5 to 18 years reporting sexual violence victimization according to the assigned sexual violence trajectory group.

https://doi.org/10.1371/journal.pone.0294207.g003

Early life factors associated with physical and sexual violence victimization trajectories

Table 2 and 3 show results of the independent and adjusted associations between the early life factors and trajectory group membership for physical and sexual violence victimization, respectively. Both individual and family level factors were associated with physical violence victimization trajectories in the unadjusted and adjusted models. In contrast, none of the family level factors associated with sexual violence victimization trajectories remained statistically significant after controlling for other variables. The high proportion of missing data on certain variables (relative growth variables, maternal prior violence experience and mode of delivery) prevented their inclusion in the multivariable model.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Univariable and multivariable logistic regression model showing early life factors associated with physical violence victimization trajectory group membership.

https://doi.org/10.1371/journal.pone.0294207.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Univariable and multivariable logistic regression model showing early life factors associated with sexual violence victimization trajectory group membership.

https://doi.org/10.1371/journal.pone.0294207.t003

Factors associated with physical violence victimization trajectories.

Of the individual level factors, sex, relative infant weight gain and childhood height gain were independently associated with membership in the chronic increasing physical violence victimization trajectory group (Table 2). From the univariable analysis, an increase in infant weight was associated with 14% greater odds of membership in the chronic increasing trajectory group (OR 1.14, 95% CI 1.02; 1.27). The results show that height gain during childhood was protective against membership in the chronic increasing trajectory group. An increase in relative height was associated with 22% lower odds of membership in the chronic increasing trajectory group (OR 0.88, 95% CI 0.79; 0.99). After adjusting for other variables in the multivariable model, sex was the only individual level factor that was significantly associated with membership in the chronic increasing trajectory group. Boys had 67% higher odds of membership in the chronic increasing trajectory group compared to girls (aOR 1.67, 95% CI 1.31; 2.10).

Household socioeconomic status was the only family level factor significantly associated with membership in the chronic increasing physical violence victimization trajectory group, in the unadjusted analysis. In addition, the association between maternal education (p = 0.061), household crowding (p = 0.079) and mode of delivery (p = 0.082) approached statistical significance with chronic increasing trajectory group membership prior to adjusting for other factors. From the multivariable model, lower household socioeconomic status and lower maternal education were the only family level factors that predicted membership in the chronic increasing trajectory group, after adjusting for other variables in the model. Children from households within the middle socioeconomic status level had 32% lower odds of membership in the chronic increasing trajectory group compared to children residing in the low household socioeconomic status level (aOR 0.68, 95% CI 0.50; 0.92). Furthermore, children whose mothers had received at least some secondary education had 73% greater odds of chronic increasing trajectory group membership compared to mothers with post-school training (aOR 1.73, 95% CI 1.08; 2.76).

Factors associated with sexual violence victimization trajectories.

Results from Table 3 demonstrate that only family level factors significantly predicted membership in the late increasing sexual violence victimization trajectory group in the univariable regression analysis (p < 0.05). Children from households with high socioeconomic status and no crowding had 42% (OR 0.58, 95% CI 0.41; 0.81) and 26% (OR 0.74, 95% CI 0.60; 0.92) lower odds of membership in the late increasing trajectory group compared to children from lower socioeconomic households and children from crowded households, respectively. Independently, the odds of membership in the late increasing trajectory group were greater with decreasing maternal (primary & below OR 2.56, 95% CI 1.55; 4.23, secondary OR 1.83, 95% CI 1.18; 2.83) and paternal (primary & below OR 1.70 95% CI 1.02; 2.83, secondary OR 1.60, 95% CI 1.19; 2.34) education compared to children from parents with post-school training. None of these factors remained statistically significant after adjusting for all other variables in the multivariable model. However, the association between household socioeconomic status and membership in the late increasing trajectory group approached significance (p = 0.07) in the multivariable analysis. Children residing in high socioeconomic households had 37% lower odds of membership in the late increasing trajectory group compared to children residing in households with low socioeconomic status (aOR 0.63, 95% CI 0.42; 0.95), after adjusting for covariates in the model.

Post regression goodness of fit analyses for the final models for the physical (p = 0.262) and sexual (p = 0.520) violence victimization trajectories, did not provide evidence of a lack of model adequacy. Low variance inflation factor (VIF) scores from the multivariable analyses – mean VIF of 1.2 for physical violence and 1.18 for sexual violence victimization – indicated no multicollinearity among covariates in the final models.

Discussion

This study sought to examine the number and forms of physical and sexual violence victimization trajectories between the ages of five and 18 years, as well as identify factors in early life associated with trajectory group membership. The results from the group-based trajectory modelling identified two trajectories of physical and two trajectories of sexual violence victimization for the best description of physical violence victimization and sexual violence victimization patterns. Overall, both the individual level – sex – and family level – household socioeconomic status and maternal education – factors were identified as important determinants of physical violence victimization trajectory group membership. However, vulnerability to increasing sexual violence victimization was largely determined by family level factors, notably household socioeconomic status.

For patterns of physical violence, the findings show that the majority of the children in the present study exhibited an increasing pattern of physical violence as they grew older and engaged with more environments outside the home. These experiences of physical violence victimization primarily occurred between early childhood and mid-adolescence. Previous literature supports these findings showing that cases of physical violence victimization peak during early adolescent years and decrease with increasing age after mid-adolescence [40–42]. However, for a third of the children in this study, their experience of physical violence did not decrease in mid-adolescence, and gradually increased as they reached adulthood. Evidence of the presence of an adolescent limited and increasing physical violence trajectory group was reported by Semenza et al. [11], among individuals assessed across four data collection points between 12 and 34 years of age. Similar to our findings, the adolescent limited trajectory group was characterized by elevated levels of violence victimization in early adolescence that decreased rapidly as respondents transitioned to adulthood. In contrast, the increasing physical violence trajectory group was characterized by low experiences of physical violence in adolescence, which then increased into adulthood. Two other trajectory groups, little to no victimization and high decreasing trajectory groups were also identified by these authors [11].

Experiences of sexual violence victimization were concentrated in the adolescent period. Parent-reported cases of sexual violence victimization were less than 1% for children at five years of age. This was followed by a steady increase in experiences of sexual violence with age during adolescence, and by the age of 18, nearly one in three children in the sample experienced some form of sexual violence. Other studies have documented a higher likelihood of sexual violence victimization among adolescent and adult samples compared to younger child samples [41,43]. For three quarters of the children, low levels of sexual violence experiences were reported at 11 years, with experience of sexual violence decreasing from 15 years of age. In contrast, a quarter of these children’s experience of sexual violence increased exponentially as they approached adulthood. Similar to our results, Jones et al. [9] generated two trajectory groups of sexual violence victimization assessed between the ages of 2–12 years. The two groups were (a) no sexual violence victimization, (b) a curvilinear pattern of sexual violence victimization characterized by low violence allegations at younger and older ages and a peak in sexual violence victimization between ages 4–8 years. This study recruited high risk children with reported alleged or substantiated allegations of child maltreatment or witnessing violence from child protective services records, which may exclude those in the population not within this radar [9].

Our findings show that certain individual level factors predicted chronic increasing physical violence victimization trajectories. Independent of other factors, boys were almost twice as likely as girls to be in the chronic increasing trajectory group. Similar findings have been demonstrated in other studies [2,11,40,44], and a possible reason for this is the fact that boys experience more conduct problems and exhibit greater externalizing behaviour compared to girls, which places them at a heightened risk for both physical violence victimization and perpetration [20,45]. Weight and height gain in early life were associated with physical violence victimization trajectories. From the univariable analyses, an increase in infant weight was a risk factor for chronic increasing physical violence trajectory group membership. In addition, being taller had a protective effect against membership in the chronic increasing physical violence victimization group. Poor linear growth (stunting) during infancy has been associated with delayed cognitive development, poor educational outcomes in childhood [29,31], increased weight gain and obesity in later childhood, adolescence and adulthood [29,46] – all factors that subsequently increase the risk of violence victimization [33]. Moreover, previous research has shown that childhood stunting results from the complex interplay between demographic, cultural, economic and environmental factors [47,48]. These same factors are linked to violence victimization, suggesting a need for further research on the association between childhood stunting and VAC, taking into account these underlying influences.

None of the individual-level factors were significantly associated with sexual violence victimization trajectories in either the crude or adjusted analyses. Although boys had a lower likelihood of membership in the late increasing sexual violence victimization group, this association was not significant. Contrary to this, many studies report significantly higher prevalence of sexual violence among girls than boys [41,43,49]. The lack of a significant difference in trajectory group membership between boys and girls may be explained by the use of self-completed questionnaires from the age of 11 years, which has been documented to contribute to higher disclosure on sexual violence experiences, especially among boys [50]. None of the infant and child growth factors were associated with late increasing sexual violence victimization group membership. Richter et al. [51] found conflicting results in the same cohort with stunted boys having a higher likelihood of sexual violence experience compared to boys with normal height for their age.

Family level factors – household socioeconomic status and maternal education – independently, and when adjusted for other confounding variables, predicted physical violence victimization trajectory group membership. Adjusting for other factors, higher household socioeconomic status was associated with a lower risk of membership in the chronic increasing physical violence victimization trajectory group. This corresponds with evidence from other studies showing that physical violence experiences are higher among poorer households [33,44]. Furthermore, maternal education was found to be protective against chronic increasing physical violence victimization trajectory group membership. Societal gender roles assign the responsibility of raising children to mothers in the household, putting further emphasis on the role of women in children’s exposure and experiences of violence. A study by Sui et al. [13], demonstrated that maternal presence in the household was protective in terms of both violence victimization and perpetration. Similar to our results, Semenza et al. [11], found that higher parental education was associated with a lower probability of membership in the increasing physical violence victimization compared to no victimization trajectory group. However, no association was observed between paternal education – a variable associated with household financial support – and chronic increasing physical violence victimization trajectories in this current study. The spillover effects of post-Apartheid migration policies on family disruption, low and delayed marriage rates and high unemployment rates have all contributed to increased rates of paternal absenteeism, non-residence or non-involvement in South African households [52]. Bitalo et al. [53] argued that the role of fathers should not be limited to the biological association or financial provision alone, rather more emphasis should be placed on the father’s or father figure’s active presence and involvement. The lack of this paternal involvement is a documented risk factor for multiple types of violence [23].

For sexual violence victimization trajectories, higher socioeconomic status appeared to be protective against inclusion in the late increasing sexual violence victimization group. However, this association was not significant after other factors were taken into account. Richter et al. [51] found similar findings, with household socioeconomic status being only significantly associated with sexual abuse among boys prior to adjusting for other covariates. Other family level factors associated with late increasing sexual violence victimization trajectories, only in the unadjusted analysis, include lower maternal and paternal education and household crowding. Consistent with our findings, Ward et al. [50], found an association that approached significance in the univariable, but not in the multivariable analysis, between sharing a bedroom with more than one person and sexual abuse. Similar to our findings, marital status, maternal education and father’s presence were not found to be significantly associated with sexual abuse among boys in South Africa [51].

South Africa’s context of high levels of all forms of violence and multiple adversities mean that many children are often raised in contexts of high adversity experienced as normative, and indeed moderate levels of adversity have been shown to contribute to resilience [54]. Further research is needed to fully understand the development of resilience in high adversity settings, and its protective role against victimization group inclusion, to expand some of the important contextualized resilience research being undertaken in South Africa. Notably, Theron [55] posits that the development of resilience in response to significant adversity must involve interplay between individuals and their social systems as an adaptive and dynamic process. And in her work, Theron finds that youth age, caregiver warmth, school resources and teacher competence all contribute to South African learners’ levels of school engagement, as a function of resilience [56]. Financial security in the home, positive parenting and the health of the caregiver have been found to be protective of children’s violence victimization [33]. Much more work needs to be done for a comprehensive understanding of the development of resilience from early childhood through adolescence, and its role along the life course in predicting children’s victimization.

Interpretation of the study findings should take into consideration several important limitations. First, the use of secondary data limited control over the timing of key measurements. In particular, violence victimization data were not collected at consistent intervals across childhood and adolescence, which may have limited the capacity to detect more nuanced or heterogeneous trajectories of violence over time. Furthermore, the dataset did not include all potentially relevant early life exposures, and completeness of data for certain variables was a concern in some cases. This possibly contributed to residual confounding, which persists despite final model demonstrating adequate fit. Second, the use of complete case analyses raises concerns about potential selection bias. Participants with missing data on key exposures or outcomes were excluded from the analysis, which may have resulted in a sample that is systematically different from those lost to follow-up or excluded due to incomplete data. Although efforts were made to compare included and excluded samples, the possibility remains that observed associations were influenced by this attrition.

Third, the reliance on caregiver- and self-reported measures of violence exposure, particularly for sensitive issues such as sexual violence, may have introduced social desirability or reporting bias. Self-completed questionnaires from age 11 likely improved disclosure, especially among boys, but underreporting—particularly in early childhood when caregivers serve as proxy respondents—remains a concern. The stigma surrounding sexual and physical abuse, as well as normative beliefs about corporal punishment, may have further influenced response patterns. Fourth, the study population was drawn from a predominantly Black, peri-urban South African cohort. While this provides critical insight into violence trajectories in a historically marginalized and under-researched context, it also limits generalizability to other population groups within South Africa and beyond. Structural factors such as legacies of apartheid spatial planning, migration, and economic exclusion shape this context in unique ways.

Lastly, although GBTM is a powerful method for identifying distinct developmental patterns, trajectory modelling is inherently exploratory and can be influenced by model specification choices, including the number of groups and polynomial orders. While the average posterior probabilities were high, indicating good classification of individuals into trajectory subgroups, the moderate entropy values suggest some uncertainty in class assignment. This uncertainty can influence the precision of risk factor associations with trajectory group membership. Future research should incorporate additional measurement waves to improve trajectory resolution, explore alternative clustering methods, and use a broader set of model adequacy indices, such as the parametric bootstrapped likelihood ratio test (BLRT) and Lo-Mendell-Rubin test (LMR-LRT), to strengthen confidence in trajectory classifications.

Despite these limitations, the study’s prospective design, large sample size, and application of life course-informed statistical methods offer several strengths. Unlike retrospective studies that are vulnerable to long recall periods, this study used repeated measures across key developmental stages, helping to minimize recall bias and better capture the timing and progression of violence experiences. Furthermore, the application of trajectory modelling contributes to an emerging literature on longitudinal patterns of violence victimization in African contexts, where evidence remains sparse. Finally, the study’s integration of both individual and family-level early life exposures aligns with the developmental origins of health and disease perspective. While DOHaD research has traditionally focused on physical health outcomes, these findings suggest that early nutritional and socioeconomic conditions may also shape vulnerability to psychosocial and interpersonal harms, such as violence victimization. This broadens the application of the DOHaD perspective to include developmental pathways of risk and resilience in the social domain, offering new insights for early intervention and prevention strategies in high-adversity settings.

Conclusion

This study demonstrates that even within shared neighbourhoods and environments, children can follow markedly different developmental patterns of violence victimization. Using group-based trajectory modelling in a longitudinal South African cohort, we identified distinct pathways of both physical and sexual violence from early childhood through adolescence. This approach offers a novel contribution to the violence literature in African contexts, where prospective studies of this kind remain rare.

The analysis revealed two key patterns of physical violence victimization: a large group experiencing increasing exposure through adolescence, and a smaller group with persistent and escalating experiences. Individual-level factors, sex and early growth indicators such as higher infant weight and shorter stature, were associated with greater risk of belonging to the chronic increasing trajectory. Family-level determinants, including lower household socioeconomic status and maternal education, also emerged as significant and consistent predictors.

In contrast, sexual violence victimization trajectories were more sharply concentrated in adolescence. While some children experienced only transient or declining exposure, a notable subgroup faced sharply increasing risk as they neared adulthood. Unlike physical violence, no individual-level factors were significantly associated with these trajectories in adjusted models, though family-level socioeconomic adversity showed some influence in unadjusted analyses.

These findings have important implications for the timing and design of early interventions. First, they highlight the value of integrated, multisectoral programmes that combine poverty alleviation with parenting and nutritional support, especially during the first 1,000 days of life. Addressing early stunting and promoting healthy development may have long-term protective effects. Second, households with low maternal education and fewer resources should be prioritized for targeted violence prevention efforts. However, the heterogeneity of trajectories also points to the need for universal, population-level strategies that can reach vulnerable children who do not fall into easily defined “high-risk” groups. A hybrid model that combines targeted interventions for chronically exposed children with broader structural initiatives may be the most effective and equitable approach, particularly in the face of persistent inequality, poverty, and systemic barriers. Community-led and contextually grounded solutions will be vital to ensuring feasibility and sustainability in this context.

In conclusion, this study contributes new evidence on developmental patterns of violence victimization and their early life origins in South Africa. It highlights the importance of longitudinal approaches for identifying risk and resilience pathways, and calls for investment in early, sustained, and layered intervention strategies that span the individual, household, and structural levels. Future research should build on this work by exploring mechanisms of resilience, refining early risk markers, and testing the effectiveness of intervention models tailored to these developmental trajectories.

Supporting information

Acknowledgments

The authors would like to thank the Birth to Twenty-Plus participants for their role in the study, as well as the Developmental Pathways to Health Research Unit at the University of the Witwatersrand where the study is housed.